Pysnmp openServerMode with IPv6 - python-2.7

Question

I try to start a SNMP Agent with pysnmp.
for IPv4 and IPv6 bind, it works pretty fine with localhost ('127.0.0.1' and '::1')
but when I try to use other IPv6 IP which I fetched from interface, it failed due to
[vagrant#test SOURCES]$ sudo python snmp_agent.py enp0s8
Traceback (most recent call last):
File "snmp_agent.py", line 172, in <module>
master_agent_startup(ifname=sys.argv[1])
File "snmp_agent.py", line 101, in master_agent_startup
(get_ipv6_address(interface_name), SNMP_AGENT_PORT))
File "/usr/lib/python2.7/site-packages/pysnmp/carrier/asyncore/dgram/base.py", line 50, in openServerMode
raise error.CarrierError('bind() for %s failed: %s' % (iface, sys.exc_info()[1],))
pysnmp.carrier.error.CarrierError: bind() for ('fe80::a00:27ff:fe9e:9c16', 8001) failed: [Errno 22] Invalid argument
This is the output from the interface 'enp0s8':
[vagrant#test SOURCES]$ ifconfig enp0s8
enp0s8: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
inet 172.20.20.26 netmask 255.255.255.0 broadcast 172.20.20.255
inet6 fe80::a00:27ff:fe9e:9c16 prefixlen 64 scopeid 0x20<link>
ether 08:00:27:9e:9c:16 txqueuelen 1000 (Ethernet)
RX packets 874053 bytes 115842841 (110.4 MiB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 862314 bytes 114652475 (109.3 MiB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
This is the code piece I used for IPv6 bind:
def get_ipv6_address(ifname):
return netifaces.ifaddresses(ifname)[netifaces.AF_INET6][0]['addr'].split('%')[0]
config.addSocketTransport(snmpEngine, udp.domainName,
udp.UdpTransport().openServerMode(
(get_ipv4_address(interface_name), SNMP_AGENT_PORT))
)
config.addSocketTransport(snmpEngine, udp6.domainName,
udp6.Udp6SocketTransport().openServerMode(
(get_ipv6_address(interface_name), SNMP_AGENT_PORT))
)
From pysnmp sample, it seems the parameter inside "openServerMode()" is just a tuple of IP and port.
And from output error I suppose there is no error for given IP and port.
so why it failed due to Invalid argument?
Could you #Ilya Etingof or some other pysnmp expert help me with it?
Thanks.
UPDATE: I try to bind it with given suggestion, but is still doesn't work.
the bind command was run from a new installed CentOS. but it still failed:
[root#test ~]# ifconfig
eth1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST> mtu 1500
inet 10.10.20.4 netmask 255.255.255.0 broadcast 10.10.20.255
inet6 fe80::f816:3eff:fee1:5475 prefixlen 64 scopeid 0x20<link>
ether fa:16:3e:e1:54:75 txqueuelen 1000 (Ethernet)
RX packets 12242 bytes 962552 (939.9 KiB)
RX errors 0 dropped 0 overruns 0 frame 0
TX packets 12196 bytes 957826 (935.3 KiB)
TX errors 0 dropped 0 overruns 0 carrier 0 collisions 0
[root#test ~]# python
Python 2.7.5 (default, Oct 11 2015, 17:47:16)
[GCC 4.8.3 20140911 (Red Hat 4.8.3-9)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import socket
>>> s = socket.socket(socket.AF_INET6, socket.SOCK_DGRAM, 0)
>>> addr_and_port = ('fe80::f816:3eff:fedb:ba4f', 8001)
>>> s.bind(addr_and_port)
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/usr/lib64/python2.7/socket.py", line 224, in meth
return getattr(self._sock,name)(*args)
socket.error: [Errno 22] Invalid argument
>>>
[1]+ Stopped python
[root#test ~]# netstat -anp | grep 8001
[root#test ~]#
One more update:
I suppose the bind is failed due to my environment has some issue with IPv6 configuration. As I'm only able to get one IPv4 address by using "socket.getaddrinfo()" method.
Br,
-Dapeng Jiao

Answer 1

You could get this error if attempted to bind the same socket more than once. But I can't see that is the case in your code.
That .openServerMode() method does no magic -- it just calls .bind() on socket object. For inspiration, does this work in at your Python prompt?
from pysnmp.carrier.asyncore.dgram import udp6
addr_and_port = ('fe80::a00:27ff:fe9e:9c16', 8001)
udp6.Udp6SocketTransport().openServerMode(addr_and_port)
or even:
import socket
s = socket.socket(socket.AF_INET6, socket.SOCK_DGRAM, 0)
addr_and_port = ('fe80::a00:27ff:fe9e:9c16', 8001)
s.bind(addr_and_port)
My hope is that tests like these may help you figuring out the problem...

Answer 2

When you use an IPv6 link-local address, you must always use the scope along with it. The link-local address is not valid without its scope, thus you will receive an Invalid argument error.
For example, instead of using fe80::a00:27ff:fe9e:9c16 you must use fe80::a00:27ff:fe9e:9c16%enp0s8.